Epicardial adipose tissue radiomics predicts VR and MACE after AMI: a prospective cohort study.

During 12-month follow-up, 29.6% of AMI patients developed ventricular remodeling (VR) and 19.9% experienced major adverse cardiovascular events (MACE).

• Single-center prospective cohort study enrolled 206 AMI patients at the Second Affiliated Hospital of Anhui Medical University.
• Enrollment period: January 2022 to June 2023.
• Follow-up duration was 12 months.
• 61 patients (29.6%) developed VR and approximately 41 patients (19.9%) experienced MACE.

EAT volume was independently associated with both ventricular remodeling and MACE after AMI.

• Statistical significance was reported at P < 0.001 for both outcomes.
• EAT features were extracted from cardiac CT using Pyradiomics software.
• Statistical analyses were performed using R, Python, and SPSS.
• EAT volume was identified as an independent predictor in multivariable modeling.

The radiomics-based Model 2 (integrating EAT radiomics with clinical parameters) showed superior predictive performance compared with Model 1 (clinical parameters alone).

• Model 2 demonstrated higher AUC and C-index values across validation folds compared to Model 1.
• Predictive models were constructed using machine learning algorithms.
• Performance was evaluated across multiple validation folds, suggesting cross-validation methodology.
• The integration of EAT radiomics features with clinical parameters drove the improvement in predictive accuracy.

Experimental studies demonstrated that EAT aggravated myocardial injury, fibrosis, and apoptosis after infarction.

• These adverse effects were partially attenuated by IL-6 neutralization, implicating IL-6 as a mechanistic mediator.
• Experimental findings provided biological support for the radiomics-based predictive models.
• Outcomes measured included myocardial injury markers, fibrosis assessment, and apoptosis quantification.
• IL-6 neutralization only partially attenuated the effects, suggesting additional mediating pathways.

EAT radiomics features were systematically extracted from cardiac CT imaging for model development.

• Pyradiomics was used as the feature extraction platform.
• The study design was prospective and single-center.
• Machine learning algorithms were employed to construct predictive models from extracted radiomics features.
• Significance threshold was set at P < 0.05 for all statistical analyses.